<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#ffffff">
Hi Ina,<br>
<br>
<blockquote
cite="mid:20110505091247.73266x7pqtrxxji8@home.staff.uni-marburg.de"
type="cite">I am refining a structure of 2.3A resolution. The R
values with and without bulk solvent and anisotropic scale are
quite different as you can see here:
<br>
<br>
Start R-work = 0.3751, R-free = 0.5562 (no bulk solvent and
anisotropic scale)
<br>
Final R-work = 0.3741, R-free = 0.5557 (no bulk solvent and
anisotropic scale)
<br>
<br>
Start R-work = 0.2298, R-free = 0.2974
<br>
Final R-work = 0.2261, R-free = 0.2970
<br>
</blockquote>
<br>
the values reported with "(no bulk solvent and anisotropic scale)"
correspond to the R-factor values calculated using
<br>
<br>
Fmodel = scale * exp(AnisoScale) * (Fcalc + k_sol * exp(-b_sol*s<sup
class="moz-txt-sup">2</sup>/4) * Fmask)
<br>
<br>
where AnisoScale = k_sol = b_sol = 0, which means "no bulk solvent
and anisotropic scale".
<br>
<br>
given a typical solvent content of macromolecular crystals ~50% and
always some anisotropy, the values do not look too surprising to me.
<br>
<br>
Actually, the R-factors before bulk-solvent correction and scaling
are rather confusing (although may be useful in some cases) so
phenix.refine does not output them anymore at the very bottom of
.log file (you can still find them somewhere in the log file).<br>
<br>
Pavel.<br>
<br>
</body>
</html>